本研究為開發即時運算以及可攜帶式的腦機介面裝置，以希爾伯特-黃轉換（Hilbert-Huang transform, HHT）為腦波分析的演算法，並且改善其運算速度的問題，將HHT演算法中的經驗模態分解法(empirical mode decomposition, EMD) 中Cubic Spline 改良成演算法計算較快的Hermite Spline，其中兩種不同形式的包絡線分別為二次及一次微分連續曲線。 並以556筆病人腦波資料進行分析，分為清醒及麻醉資料，並以希爾伯特黃平均區域性頻率(Hilbert-Huang average regional frequency, HHARF)作為判斷兩種曲線演算法相關係數之依據，其中以EMD分解出的本質模態函數(intrinsic mode functions, IMF)，IMF1至IMF6皆能以HHARF判斷麻醉及清醒，其中以IMF1更為明顯。我們再以兩種曲線演算法所計算的IMF1之HHARF，兩者之間的相關係數清醒為0.97，麻醉為0.92。 ;In this papper, we developed a portable and real time brain-computer interface device using Hilbert-Huang Transform (HHT) to analyze brain waves. In order to improve the speed of the calculation, the cubic spline of the empirical mode deposition (EMD) method in the algorithms of HHT is changed into Hermite spline. The two different forms of the envelope are quadratic differential continuous curve and a differential continuous curve, respectively. With 556 strokes of EEG data of patient were analyzed, divided awake and anesthesia information. The correlation coefficients of two kinds of algorithms are judged by using Hilbert-Huang average regional frequency (HHARF). For the intrinsic mode functions (IMF) decomposed by EMD, we can use HHARF to judge awake and anesthesia from IMF1 to IMF6. Furthermore, the result is more apparent in IMF1. Using two curves algorithm to calculate HHARF of IMF1, the correlation coefficient of awake is 0.97, and the correlation coefficient in anesthesia is 0.92.
